CN114637843A

CN114637843A - Data processing method and device, electronic equipment and storage medium

Info

Publication number: CN114637843A
Application number: CN202011482703.4A
Authority: CN
Inventors: 陈谦
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2020-12-15
Filing date: 2020-12-15
Publication date: 2022-06-17

Abstract

The embodiment of the application provides a data processing method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence. The data processing method comprises the following steps: pre-training the text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model; acquiring label-free text data and a pseudo label of the label-free text data obtained after processing the label-free text data; updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model; and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model. By the embodiment of the application, the accuracy of the text post-processing model can be higher.

Description

Data processing method and device, electronic equipment and storage medium

Technical Field

The embodiment of the application relates to the technical field of artificial intelligence, in particular to a data processing method and device, electronic equipment and a storage medium.

Background

Text post-processing is a process of post-processing text obtained based on automatic speech recognition techniques. In general, text post-processing includes: the method comprises two tasks of punctuation prediction and smooth detection, and can establish a text post-processing model in a multi-task learning mode, thereby realizing the text post-processing operation of the text to be processed.

At present, supervised text data is mainly used for training a text post-processing model. Specifically, the method comprises the following steps: firstly, acquiring a large amount of standard text data from a standard library (such as a library of Wikipedia and the like) as training labels, generating training samples based on the standard text data, and then training an initial text post-processing model according to the model training samples and the labels to obtain a corresponding model.

The above process uses a large amount of standard text data for model training. However, because the standard text data are generally single in source and limited in related application fields, the accuracy of the text post-processing model obtained by the training method is low.

Disclosure of Invention

The present application is directed to a data processing method, an apparatus, an electronic device, and a computer storage medium, which at least partially solve the problems in the prior art.

According to a first aspect of embodiments of the present application, there is provided a data processing method, including:

pre-training the text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model;

acquiring label-free text data and a pseudo label of the label-free text data obtained after processing the label-free text data;

training and updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model;

and training and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model.

According to a second aspect of embodiments of the present application, there is provided a data processing method, including:

acquiring text data to be processed;

inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a third aspect of embodiments of the present application, there is provided a data processing method, including:

receiving an instruction which is input through an interface of the instant messaging application and used for indicating that input voice data are converted into text data;

performing text conversion on the voice data according to the instruction to obtain text data to be processed;

inputting the text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a fourth aspect of the embodiments of the present application, there is provided a data processing method, including:

receiving an instruction which is input and set through the all-in-one machine equipment and is used for indicating that the input voice data is converted into text data;

According to a fifth aspect of embodiments of the present application, there is provided a data processing method, including:

receiving voice data uploaded by a public cloud client;

performing text conversion on the voice data to obtain text data to be processed;

According to a sixth aspect of the embodiments of the present application, there is provided a data processing method, including:

receiving to-be-processed text data uploaded by a public cloud client, wherein the to-be-processed text data is obtained after the public cloud client performs text conversion on received voice data;

According to a seventh aspect of embodiments of the present application, there is provided a data processing apparatus. The device comprises:

the model pre-training module is used for pre-training the text post-processing model by adopting the first text data training sample and the label corresponding to the first text data training sample to obtain a pre-training model;

the non-label text data and pseudo label acquisition module is used for acquiring non-label text data and pseudo labels of the non-label text data obtained after the non-label text data is processed;

the first training updating module is used for training and updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model;

and the second training updating module is used for training and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model.

According to an eighth aspect of embodiments of the present application, there is provided a data processing apparatus. The device comprises:

the text data to be processed acquisition module is used for acquiring text data to be processed;

the first processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring the processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a ninth aspect of an embodiment of the present application, there is provided a data processing apparatus. The device comprises:

the first instruction receiving module is used for receiving an instruction which is input through an interface of the instant messaging application and used for indicating that input voice data are converted into text data;

the first text conversion module is used for performing text conversion on the voice data according to the instruction to obtain text data to be processed;

the second processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a tenth aspect of the embodiments of the present application, there is provided a data processing apparatus. The device comprises:

the second instruction receiving module is used for receiving an instruction which is input and set through the all-in-one machine equipment and is used for indicating to convert the input voice data into text data;

the second text conversion module is used for performing text conversion on the voice data according to the instruction to obtain text data to be processed;

the third processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to an eleventh aspect of embodiments of the present application, there is provided a data processing apparatus. The device comprises:

the voice data receiving module is used for receiving voice data uploaded by the public cloud client;

the third text conversion module is used for performing text conversion on the voice data to obtain text data to be processed;

the fourth processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a twelfth aspect of embodiments of the present application, there is provided a data processing apparatus. The device comprises:

the system comprises a to-be-processed text data receiving module, a to-be-processed text data processing module and a text conversion module, wherein the to-be-processed text data is obtained by performing text conversion on received voice data by a public cloud client;

a fifth processed text data obtaining module, configured to input the text data to be processed into a text post-processing model, and obtain processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of the first aspect.

According to a thirteenth aspect of embodiments of the present application, there is provided an electronic apparatus including: one or more processors; a computer readable medium configured to store one or more programs which, when executed by one or more processors, cause the one or more processors to implement the data processing method of any one of the first to sixth aspects of the embodiments described above.

According to a fourteenth aspect of embodiments of the present application, there is provided a computer-readable medium on which a computer program is stored, the program, when executed by a processor, implementing the data processing method of any one of the first to sixth aspects of the embodiments described above.

According to a fifteenth aspect of embodiments herein, there is provided a computer program containing computer-executable instructions that, when executed, implement the data processing method of any one of the first to sixth aspects of the embodiments described above.

According to the data processing scheme provided by the embodiment of the application, after the pre-training model is obtained by using the first text data training sample and the label corresponding to the first text data training sample, the pre-training model is not used as a final text post-processing model. And updating the pre-training model based on the unlabeled text data and the corresponding pseudo labels to obtain an updated model, and then updating the updated model again by using the second text data training sample and the labels corresponding to the second text data training sample to obtain the trained text post-processing model. That is to say, compared with the conventional model training process, in the scheme provided in the embodiment of the present application, the pre-training model is updated twice, so that the trained text post-processing model is obtained. Therefore, the accuracy of the text post-processing model obtained by the data processing method provided by the embodiment of the application is higher.

Drawings

Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:

FIG. 1a is a flowchart illustrating steps of a data processing method according to an embodiment of the present disclosure;

FIG. 1b is a schematic diagram of a data processing flow according to an embodiment of the present application;

FIG. 2 is a flowchart illustrating steps of a data processing method according to a second embodiment of the present application;

FIG. 3a is a flowchart illustrating steps of a data processing method according to a third embodiment of the present application;

FIG. 3b is a schematic diagram of a data processing flow according to the third embodiment of the present application;

FIG. 4 is a flowchart illustrating steps of a data processing method according to a fourth embodiment of the present application;

FIG. 5 is a flowchart illustrating steps of a data processing method according to a fifth embodiment of the present application;

FIG. 6 is a flowchart illustrating steps of a data processing method according to a sixth embodiment of the present application;

FIG. 7 is a flowchart illustrating steps of a data processing method according to a seventh embodiment of the present application;

FIG. 8 is a schematic structural diagram of a data processing apparatus according to an eighth embodiment of the present application;

FIG. 9 is a schematic structural diagram of a data processing apparatus according to a ninth embodiment of the present application;

FIG. 10 is a schematic structural diagram of a data processing apparatus according to a tenth embodiment of the present application;

FIG. 11 is a schematic structural diagram of a data processing apparatus according to an eleventh embodiment of the present application;

FIG. 12 is a schematic structural diagram of a data processing apparatus according to a twelfth embodiment of the present application;

FIG. 13 is a schematic diagram of a data processing apparatus in an embodiment thirteen in the present application;

fig. 14 is a schematic structural diagram of an electronic device in a fourteenth embodiment of the application;

fig. 15 is a hardware structure of an electronic device in fifteenth embodiment of the present application.

Detailed Description

The present application will be described in further detail with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the related invention are shown in the drawings.

It should be noted that, in the present application, the embodiments and features of the embodiments may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.

Referring to fig. 1a, a flowchart of steps of a data processing method according to a first embodiment of the present application is shown.

Specifically, the data processing method provided by this embodiment includes the following steps:

step 101, pre-training a text post-processing model by using a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model.

In this step, the first text data training sample may be any text data obtained based on an automatic speech recognition technique. For example, using the written corpora (standard text data) in the existing written corpus (standard text database) to generate the non-smooth corpora, so that the generated non-smooth corpora is used as the first text data training sample, and the corresponding written corpora can be the label.

The text post-processing model in the embodiment of the present invention may be any deep learning model, for example: a convolutional neural network model, a cyclic neural network model, and the like, and here, the specific form of the text postprocessing model is not limited.

Usually, the number of training samples is huge, and if the labels corresponding to the training samples are obtained in a manual labeling mode, the cost is high. Thus, in some alternative embodiments, prior to performing this step, one may: acquiring standard text data from a standard text database, and generating corresponding non-smooth text data by adopting a preset rule; and taking the non-smooth text data as a first text data training sample, and taking the standard text data as a label corresponding to the first text data training sample. Compared with a manual labeling mode, the mode can reduce labor cost.

The non-smooth text data means repeated words or redundant tone-assisted words or text data with unsmooth semantics. The preset rule may be set by a person skilled in the art as appropriate according to actual requirements, and may be, for example: punctuation marks in the standard text data are filtered, and preset Chinese auxiliary words (such as kaempferia, o and the like) are randomly added in the filtered standard text data; the following steps can be also included: punctuation marks in the standard text data are filtered, and partial words and the like in the standard text data after filtering are randomly repeated. In the embodiment of the present application, specific contents of the preset rule may be set according to an actual situation, and here, the specific contents are not limited.

And 102, acquiring the unlabeled text data and a pseudo label of the unlabeled text data obtained after the unlabeled text data is processed.

In the embodiment of the application, the pseudo label of the unlabeled text data can be obtained by performing text post-processing pretreatment on the unlabeled text data. Since the model is not trained at this point, the obtained labels may not be accurate enough and are therefore referred to as pseudo labels.

In some optional embodiments, when obtaining the non-tag text data and the pseudo tag of the non-tag text data obtained by processing the non-tag text data, the following steps may be performed: acquiring label-free text data; and performing label prediction on the label-free text data by adopting a label prediction model to obtain a pseudo label of the label-free text data.

Alternatively, the unlabeled text data may be acquired as follows:

acquiring voice data to be recognized; and adopting an automatic voice recognition technology to recognize voice data to be recognized to obtain non-tag text data.

According to the method, the voice data to be recognized is obtained firstly, and then the non-tag text data is obtained by means of an automatic voice recognition technology, so that a large amount of non-tag text data can be obtained quickly, and the method can be effectively applied to ASR scenes.

Optionally, the label prediction model may be obtained in the following two manners:

first, the model may be obtained after performing training update based on the pre-training model obtained in step 101. Specifically, the text data training sample and the label corresponding to the text data training sample may be obtained again, and then the network parameters in the pre-training model obtained in the step 101 are trained and updated based on the text data training sample and the label corresponding to the text data training sample obtained again, so as to obtain the label prediction model.

The text data training sample obtained again can be text data to be post-processed with a high matching degree with the application field (target field) of the text post-processing model, and correspondingly, in order to ensure the accuracy of the obtained label, the text post-processing can be performed on the text data training sample obtained again in a manual participation mode, so as to obtain the label corresponding to the text data training sample.

Second, the model may be obtained after performing training update based on another model having a larger scale and higher accuracy than the pre-training model obtained in step 101.

For example, an initial label prediction model may be constructed first, where the number of network layers included in the label prediction model is greater than the number of network layers included in the pre-training model, or the dimension of each network layer in the label prediction model is greater than the dimension of each network layer in the pre-training model; then, the corpus in the existing corpus is used as a training sample, and the initial label prediction model is pre-trained to obtain a pre-trained label prediction model; and then acquiring a text data training sample and a label thereof which are highly matched with the application field (target field) of the text post-processing model, training and updating the network parameters of the label prediction model after pre-training, and finally obtaining the label prediction model after training. The label prediction model has a larger number of network layers or dimension of each network layer, so that the accuracy rate of the label prediction model is higher.

Compared with the two label prediction models in the acquisition modes: the first mode is that training and updating are carried out on the basis of the pre-training model obtained in the step 101, so that the implementation process is simpler, and the acquisition speed of the label prediction model is higher; the second method is not based on the pre-training model obtained in step 101, but based on another model which is larger in scale and has higher accuracy than the pre-training model obtained in step 101, and is obtained by performing training and updating, so that the training process of the model is complicated, the tag prediction model is slow to obtain, but the accuracy of the obtained tag prediction model is higher.

Specifically, in some optional embodiments, performing label prediction on the unlabeled text data by using a label prediction model to obtain a pseudo label of the unlabeled text data may include:

training and updating the pre-training model based on the third text data training sample and a label corresponding to the third text data training sample to obtain a label prediction model; and performing label prediction on the label-free text data by adopting a label prediction model to obtain a pseudo label of the label-free text data.

The third text data training sample may be: correspondingly, in order to ensure the accuracy of the obtained label, the text post-processing can be performed on the third text data training sample in a manual participation mode, so that the label corresponding to the third text data training sample is obtained.

For example, the third text data training sample and the label thereof may be: and (3) manually labeling the corpora in the spoken language corpus in a smaller scale. Specifically, the corpus in the artificially labeled spoken language corpus can be used as a third text data training sample, and the corresponding artificially labeled content is used as a label.

In other alternative embodiments, performing label prediction on the unlabeled text data by using a label prediction model to obtain a pseudo label of the unlabeled text data may include:

acquiring a pre-constructed initial label prediction model; the number of network layers contained in the label prediction model is more than that contained in the text post-processing model, and/or the dimensionality of each network layer in the label prediction model is more than that of each network layer in the text post-processing model;

pre-training the initial label prediction model by adopting a fourth text data training sample and a label corresponding to the fourth text data training sample to obtain a pre-trained label prediction model;

training and updating the pre-trained label prediction model based on a fifth text data training sample and a label corresponding to the fifth text data training sample to obtain a trained label prediction model;

and performing label prediction on the label-free text data by adopting the trained label prediction model to obtain a pseudo label of the label-free text data.

As for the fourth text data training sample, since the function of the fourth text data training sample is the same as that of the first text data training sample in step 101, and the fourth text data training sample is used for model pre-training, in some optional embodiments, the written corpora in the existing written corpus may also be used to generate the non-smooth corpora, so that the generated non-smooth corpora is used as the fourth text data training sample, and the written corpora corresponding to the fourth text data training sample may be tags.

For the fifth text data training sample, the same function as that of the third text data training sample is used for training and updating the model, so in some optional embodiments, a corpus in a smaller-scale artificial labeled spoken language corpus may also be used as the fifth text data training sample, and the corresponding artificial labeled content is used as a label.

And 103, training and updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model.

The pre-training model is obtained by pre-training based on corpora in the existing corpus, and the accuracy of the pre-training model is low due to the fact that the application field related to the existing corpus is limited and the matching degree between the pre-training model and the application field of the text post-processing model is not high. Based on the reasons, a large amount of label-free text data (such as ASR manual transcription text) related to more application fields can be obtained after the pre-training model is obtained, and then the network parameters of the pre-training model are trained and updated based on the label-free text data and the pseudo labels thereof, so that the updated model with higher accuracy can be obtained.

And 104, training and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model.

Similar to the third text data training sample and the fifth text data training sample, the second text data training sample in this step may be used as a sample for performing model training and updating, and may be: correspondingly, in order to ensure the accuracy of the obtained label, the text post-processing can be performed on the fifth text data training sample in a manual participation mode, and then the label corresponding to the fifth text data training sample is obtained. For example, in some optional embodiments, a corpus in a smaller-scale artificial labeled spoken language corpus may also be used as a second text data training sample, and the corresponding artificial labeled content may also be used as a label.

According to the data processing method, the data processing device, the electronic equipment and the storage medium provided by the embodiment of the application, the data processing method comprises the following steps: pre-training the text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model; acquiring label-free text data and a pseudo label of the label-free text data obtained after processing the label-free text data; updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model; and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model.

In the embodiment of the application, after the pre-training model is obtained by using the first text data training sample and the label corresponding to the first text data training sample, the pre-training model is not used as a final text post-processing model. And updating the pre-training model based on the label-free text data and the corresponding pseudo label to obtain an updated model, and then updating the updated model again by using the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model. That is to say, compared with a conventional model training process, in the model training method provided in the embodiment of the present application, the pre-training model is updated twice, so that a trained text post-processing model is obtained. Therefore, the accuracy of the text post-processing model obtained by the data processing method provided by the embodiment of the application is higher.

The data processing method provided by the embodiment of the present application may be executed by any suitable device with data processing capability, including but not limited to: a terminal, a mobile terminal, a PC, a server and the like.

Referring to fig. 1b, fig. 1b is a schematic diagram of a data processing flow according to an embodiment of the present application. The following briefly describes a data processing flow provided by the first embodiment of the present application with reference to fig. 1b, which mainly includes:

after the initial text is constructed, the model is processed, for example: after the initial Transformer model is constructed,

the first step is as follows: and pre-training the constructed initial text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model. Specifically, the method comprises the following steps: the initial transform model may be pre-trained using corpora in a large-scale existing written corpus, where non-smooth data may be generated for the corpus using a preset rule as a first text data training sample, and the corpora corresponding to the non-smooth data may be used as a label.

The second step: and acquiring a third text data training sample and a sample corresponding to the third text data training sample, and training and updating the pre-training model to obtain a prediction model. Specifically, the corpus in the manually labeled spoken language corpus with a smaller scale may be used as the third text data training sample, and the manually labeled content corresponding to the third text data training sample may be used as the label.

The third step: the model self-training is carried out by adopting label-free text data obtained by an automatic speech recognition technology, and specifically comprises the following steps: performing label prediction on the label-free text data obtained by the automatic speech recognition technology by using the prediction model to obtain a pseudo label of the label-free text data; training and updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model;

the fourth step: and training and updating the updated model based on the second text data training sample and the corresponding label to obtain a trained text post-processing model. Specifically, the corpus in the manually labeled spoken language corpus with a smaller scale may be used as the second text data training sample, and the manually labeled content corresponding to the second text data training sample may be used as the label.

Referring to fig. 2, a flowchart illustrating steps of a data processing method according to a second embodiment of the present application is shown.

Specifically, the data processing method provided by the embodiment of the present application includes the following steps:

step 201, text data to be processed is obtained.

Step 202, inputting the text data to be processed into a text post-processing model, and acquiring the processed text data output by the text post-processing model.

The text post-processing model may be obtained based on the data processing method in the first embodiment, and is not described herein again.

In the embodiment of the application, after the text data to be processed is obtained, the text data to be processed is input to the text post-processing model obtained based on the data processing method in the first embodiment, and then the processed text data output by the text post-processing model is obtained.

In the first embodiment, in the data processing process of the text post-processing model, the first text data training sample and the label corresponding to the first text data training sample are used, so that the pre-training model is not used as the final text post-processing model after being obtained. And updating the pre-training model based on the unlabeled text data and the corresponding pseudo labels to obtain an updated model, and then updating the updated model again by using the second text data training sample and the labels corresponding to the second text data training sample to obtain the trained text post-processing model. That is to say, compared with the conventional model training process, in the data processing method provided in the first embodiment, the pre-training model is updated twice, so that the trained text post-processing model is obtained. Therefore, the accuracy of the text post-processing model obtained by the data processing method provided by the first embodiment is higher.

Further, the text data to be processed is input into the text post-processing model obtained by using the data processing method provided in the first embodiment, so that the processed text data with higher accuracy can be obtained.

Referring to fig. 3a, a flowchart of steps of a data processing method according to a third embodiment of the present application is shown.

step 301, acquiring text data of online log reflow as text data to be processed.

The text data of the online log reflow is an actual processing object of the text post-processing model in the application stage, that is, the text data of the online log reflow relates to the application field, namely the target field of the text post-processing model application.

Therefore, the online log reflow pseudo label is predicted through the text post-processing model as the text data to be processed, then the text data and the online log reflow pseudo label are adopted as the training samples to train and update (fine tune) the model, and the accuracy of the text post-processing model can be continuously improved along with the increase of the online log reflow text data.

Step 302, inputting the text data to be processed into the text post-processing model, and acquiring the processed text data output by the text post-processing model.

The text post-processing model may be obtained based on the data processing method in the first embodiment.

Inputting the text data of the online log reflow into a text post-processing model, and performing label prediction on the text data of the online log reflow through the model to obtain pseudo label data of the online log reflow, namely the processed text data output by the text post-processing model in the step.

And 303, training and updating the updated model based on the text data to be processed and the processed text data to obtain a transition model.

The updated model is the updated model in the first embodiment.

In this step, the updated model in the first embodiment is refined by using the text data and the pseudo tag thereof reflowed by the online log, that is: and training and updating the network parameters of the model to obtain a transition model.

And 304, training and updating the transition model based on the sixth text data training sample and the label corresponding to the sixth text data training sample to obtain the hot standby model.

The sixth text data training sample in this step is used as a sample for performing model training update, and may also be: correspondingly, in order to ensure the accuracy of the obtained label, text post-processing can be performed on the sixth text data training sample in a manual participation mode, and then the label corresponding to the fifth text data training sample is obtained.

For example, a corpus in a small-scale artificial labeled spoken language corpus may be used as a fifth text data training sample, and a specific text content corresponding to the fifth text data training sample may be used as a label.

In this step, the small-scale manually labeled spoken language corpus data is used to perform model refinement on the transition model obtained in step 303, that is, the network parameters of the transition model are trained and updated to obtain the hot standby model.

And 305, respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model.

In some alternative embodiments, the accuracy of the hot standby model and the text post-processing model may be calculated as follows:

acquiring a seventh text data training sample and a label corresponding to the seventh text data training sample; and respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model based on the seventh text data training sample and the label corresponding to the seventh text data training sample.

The seventh text data training sample in this step is used as a sample for model accuracy verification, and may be: and the text data to be post-processed has higher matching degree with the application field (target field) of the text post-processing model. Correspondingly, text post-processing can be performed on the seventh text data training sample in a manual participation mode, and then a label corresponding to the seventh text data training sample is obtained. For example, corpora in a small-scale artificial labeled spoken language corpus can be used as a seventh text data training sample, and the corresponding specific text content can be used as a label.

And step 306, when the accuracy of the text post-processing model is lower than that of the hot standby model, adopting the hot standby model as a new text post-processing model to perform the next text post-processing operation.

And 307, when the accuracy of the hot standby model is lower than that of the text post-processing model, performing the next text post-processing operation by using the text post-processing model.

In the embodiment of the application, after the text data of the online log reflow is acquired and used as the text data to be processed, the text data to be processed is input to the text post-processing model obtained based on the data processing method in the first embodiment, and the processed text data output by the text post-processing model is obtained. Training and updating the updated model in the first embodiment based on the text data to be processed and the processed text data to obtain a transition model; training and updating the transition model based on a sixth text data training sample and a label corresponding to the sixth text data training sample to obtain a hot standby model; respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model; and when the accuracy of the text post-processing model is lower than that of the hot standby model, the hot standby model is adopted as a new text post-processing model to carry out the next text post-processing operation.

Furthermore, the text data to be processed is input into the text post-processing model obtained by using the data processing method provided in the first embodiment, so that the processed text data with higher accuracy can be obtained.

In addition, in the third embodiment of the present application, the text data of the online log reflow is used as the text data to be processed, and then the updated model obtained in the first embodiment is trained and updated based on the text data to be processed and the processed text data corresponding to the text data to be processed, so as to obtain the transition model; training and updating the transition model to obtain a hot standby model; and respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model, and taking the model with high accuracy as the text post-processing model used when the next text post-processing operation is carried out. Therefore, the text post-processing model can be continuously updated through the text data reflowed through the online log, and the accuracy of the text post-processing model is further improved.

Referring to fig. 3b, fig. 3b is a schematic diagram of a data processing flow according to the third embodiment of the present application.

The following briefly describes a data processing flow provided by the third embodiment of the present application with reference to fig. 3b, which mainly includes:

after the text post-processing model is obtained,

the first step is as follows: adopting the text data of on-line log reflux to perform self-training on the text post-processing model, specifically: inputting the label-free text data of the online log reflux into a text post-processing model to obtain an online log reflux text data pseudo label (processing the post-processing text data), and training and updating the updated model (obtained in the first embodiment) based on the text data of the online log reflux and the online log reflux text data pseudo label to obtain a transition model;

the second step is that: training and updating the transition model based on a sixth text data training sample and a label thereof to obtain a hot standby model;

the third step: and detecting whether the accuracy of the hot standby model is higher than that of the text post-processing model in the first step, if so, replacing the text post-processing model in the first step with the hot standby model to serve as a new text post-processing model to perform the next text post-processing operation.

Referring to fig. 4, a flowchart illustrating steps of a data processing method according to a fourth embodiment of the present application is shown. The application scenario of this embodiment may be: and performing character conversion on the instant messaging voice data in the instant messaging application, and performing post-processing on the text data obtained by conversion.

step 401, an instruction for instructing to convert input voice data into text data, which is input through an interface of an instant messaging application, is received.

And 402, performing text conversion on the voice data according to the instruction to obtain text data to be processed.

Specifically, an automatic speech recognition technology may be adopted to perform text conversion on the input speech data, thereby obtaining text data to be processed.

Step 403, inputting the text data to be processed into a text post-processing model, and acquiring the processed text data output by the text post-processing model.

In the embodiment of the application, after the text data is subjected to text conversion to obtain the text data to be processed, the text data to be processed is input to the text post-processing model obtained based on the data processing method in the first embodiment, and then the processed text data output by the text post-processing model is obtained.

Referring to fig. 5, a flowchart illustrating steps of a data processing method according to a fifth embodiment of the present application is shown. The application scenario of this embodiment may be: and performing character conversion on voice data input through the all-in-one machine equipment, and performing post-processing on text data obtained by conversion.

step 501, receiving an instruction which is input and set through the all-in-one machine and used for indicating to convert input voice data into text data.

Step 502, performing text conversion on the voice data according to the instruction to obtain text data to be processed.

Step 503, inputting the text data to be processed into the text post-processing model, and acquiring the processed text data output by the text post-processing model.

In the first embodiment, in the data processing process of the text post-processing model, the first text data training sample and the label corresponding to the first text data training sample are used, so that the pre-training model is not used as the final text post-processing model after being obtained. And updating the pre-training model based on the label-free text data and the corresponding pseudo label to obtain an updated model, and then updating the updated model again by using the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model. That is to say, compared with the conventional model training process, in the data processing method provided in the first embodiment, the pre-training model is updated twice, so that the trained text post-processing model is obtained. Therefore, the accuracy of the text post-processing model obtained by the data processing method provided by the first embodiment is higher.

Referring to fig. 6, a flowchart illustrating steps of a data processing method according to a sixth embodiment of the present application is shown. The application scenario of this embodiment may be: and uploading the voice data to a cloud server by a client in the public cloud, performing text conversion by the cloud server, and performing post-processing on the text data obtained by conversion.

step 601, receiving voice data uploaded by a public cloud client.

Step 602, performing text conversion on the voice data to obtain text data to be processed.

Specifically, an automatic speech recognition technology may be adopted to perform text conversion on the received speech data, thereby obtaining text data to be processed.

Step 603, inputting the text data to be processed into a text post-processing model, and acquiring the processed text data output by the text post-processing model.

Further, the cloud server acquires the processed data, and can return the processed data to the public cloud client.

In the embodiment of the application, after the text conversion is performed on the voice data by the cloud server to obtain the text data to be processed, the text data to be processed is input to the text post-processing model obtained based on the data processing method in the first embodiment, and the processed text data output by the text post-processing model is further obtained.

Referring to fig. 7, a flowchart illustrating steps of a data processing method according to a seventh embodiment of the present application is shown. The application scenario of the embodiment may still be a public cloud scenario, and specifically may be: the method comprises the steps that a client in the public cloud carries out text conversion on received voice data, then the text data obtained through conversion is uploaded to a cloud server, text conversion is carried out through the cloud server, and post-processing is carried out on the text data obtained through conversion.

and step 701, receiving text data to be processed uploaded by a public cloud client.

The text data to be processed is obtained after the public cloud client performs text conversion on the received voice data. Specifically, an automatic speech recognition technology may be adopted to perform text conversion on the received speech data, thereby obtaining text data to be processed.

Step 702, inputting the text data to be processed into the text post-processing model, and acquiring the processed text data output by the text post-processing model.

In the embodiment of the application, after receiving the text data to be processed, the cloud server inputs the text data to be processed into the text post-processing model obtained based on the data processing method in the first embodiment, and further obtains the processed text data output by the text post-processing model.

Referring to fig. 8, a schematic structural diagram of a data processing apparatus in an eighth embodiment of the present application is shown.

The data processing device provided by the embodiment of the application comprises:

the model pre-training module 801 is configured to pre-train the text post-processing model by using the first text data training sample and the label corresponding to the first text data training sample to obtain a pre-training model;

a non-tag text data and pseudo tag obtaining module 802, configured to obtain non-tag text data and a pseudo tag of the non-tag text data obtained by processing the non-tag text data;

a first training update module 803, configured to perform training update on the pre-training model based on the unlabeled text data and the pseudo label, to obtain an updated model;

and the second training updating module 804 is configured to perform training and updating on the updated model based on the second text data training sample and the label corresponding to the second text data training sample, so as to obtain a trained text post-processing model.

Optionally, the apparatus in the embodiment of the present application further includes:

the standard text data and non-smooth text data acquisition module is used for acquiring standard text data from a standard text database and generating corresponding non-smooth text data by adopting a preset rule; and taking the non-smooth text data as a first text data training sample, and taking the standard text data as a label corresponding to the first text data training sample.

Optionally, the module 802 for acquiring unlabeled text data and pseudo labels includes:

the label-free text data unit is used for acquiring label-free text data;

and the pseudo label obtaining unit is used for adopting a label prediction model to carry out label prediction on the non-label text data to obtain a pseudo label of the non-label text data.

Optionally, the unlabeled text data unit is specifically configured to:

acquiring voice data to be recognized;

and recognizing the voice data to be recognized by adopting an automatic voice recognition technology to obtain the non-tag text data.

Optionally, the pseudo tag obtaining unit is specifically configured to:

training and updating the pre-training model based on the third text data training sample and a label corresponding to the third text data training sample to obtain a label prediction model;

and performing label prediction on the label-free text data by adopting a label prediction model to obtain a pseudo label of the label-free text data.

Optionally, the pseudo tag obtaining unit is specifically configured to:

The data processing apparatus of the embodiment of the present application is used to implement the corresponding data processing method in the first embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, the functional implementation of each module in the data processing apparatus of this embodiment can refer to the description of the corresponding part in the first method embodiment, and is not repeated herein.

Referring to fig. 9, a schematic structural diagram of a data processing apparatus in a ninth embodiment of the present application is shown.

a to-be-processed text data obtaining module 901, configured to obtain to-be-processed text data;

a first processed text data obtaining module 902, configured to input text data to be processed into a text post-processing model, and obtain processed text data output by the text post-processing model; the text post-processing model is obtained based on the data processing method of the first embodiment.

Optionally, the to-be-processed text data obtaining module 901 is specifically configured to: acquiring text data of online log reflux as text data to be processed;

the device of the embodiment of the application further comprises:

the transition model obtaining module is used for training and updating the updated model in the first embodiment based on the text data to be processed and the processed text data after the processed text data output by the text post-processing model is obtained, so as to obtain a transition model;

the hot standby model obtaining module is used for training and updating the transition model based on a sixth text data training sample and a label corresponding to the sixth text data training sample to obtain a hot standby model;

the accuracy calculation module is used for respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model;

and the text post-processing model updating module is used for adopting the hot standby model as a new text post-processing model to perform the next text post-processing operation when the accuracy of the text post-processing model is lower than that of the hot standby model.

Optionally, the apparatus in this embodiment of the present application further includes:

and the text post-processing model retaining module is used for adopting the text post-processing model to perform the next text post-processing operation when the accuracy of the hot standby model is lower than that of the text post-processing model.

Optionally, the accuracy calculation module is specifically configured to:

acquiring a seventh text data training sample and a label corresponding to the seventh text data training sample;

and respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model based on the seventh text data training sample and the label corresponding to the seventh text data training sample.

The data processing apparatus in the embodiment of the present application is configured to implement the corresponding data processing method in the second method embodiment or the third method embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, for the functional implementation of each module in the data processing apparatus in the embodiment of the present application, reference may be made to the description of the corresponding part in the foregoing second method embodiment or third method embodiment, and further description is omitted here.

Referring to fig. 10, a schematic structural diagram of a data processing apparatus in a tenth embodiment of the present application is shown.

a first instruction receiving module 1001, configured to receive an instruction input through an interface of an instant messaging application, where the instruction is used to instruct to convert input voice data into text data;

the first text conversion module 1002 is configured to perform text conversion on the voice data according to the instruction to obtain text data to be processed;

a second processed text data obtaining module 1003, configured to input text data to be processed into the text post-processing model, and obtain processed text data output by the text post-processing model; the text post-processing model is obtained based on the data processing method of the first embodiment.

The data processing apparatus in the embodiment of the present application is configured to implement the corresponding data processing method in the fourth method embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, the functional implementation of each module in the data processing apparatus in the embodiment of the present application can refer to the description of the corresponding part in the fourth method embodiment, and is not repeated here.

Referring to fig. 11, a schematic structural diagram of a data processing apparatus in an eleventh embodiment of the present application is shown.

a second instruction receiving module 1101, configured to receive an instruction which is input and set by the all-in-one machine device and is used for instructing to convert input voice data into text data;

the second text conversion module 1102 is configured to perform text conversion on the voice data according to the instruction to obtain text data to be processed;

a third processed text data obtaining module 1103, configured to input text data to be processed into a text post-processing model, and obtain processed text data output by the text post-processing model; the text post-processing model is obtained based on the data processing method of the first embodiment.

The data processing apparatus according to the embodiment of the present application is configured to implement the corresponding data processing method in the fifth method embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, the functional implementation of each module in the data processing apparatus in the embodiment of the present application can refer to the description of the corresponding part in the fifth method embodiment, and is not repeated here.

Referring to fig. 12, a schematic structural diagram of a data processing apparatus in a twelfth embodiment of the present application is shown.

the voice data receiving module 1201 is used for receiving voice data uploaded by a public cloud client;

a third text conversion module 1202, configured to perform text conversion on the voice data to obtain to-be-processed text data;

a fourth processed text data obtaining module 1203, which inputs the text data to be processed into a text post-processing model, and obtains processed text data output by the text post-processing model; the text post-processing model is obtained based on the data processing method of the first embodiment.

Optionally, the apparatus according to the embodiment of the present application may further include:

and the first processed postscript data returning module is used for returning the processed postscript data to the public cloud client.

The data processing apparatus according to the embodiment of the present application is configured to implement the corresponding data processing method in the sixth embodiment of the foregoing method, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, the functional implementation of each module in the data processing apparatus in the embodiment of the present application can refer to the description of the corresponding part in the sixth method embodiment, and is not repeated here.

Referring to fig. 13, a schematic structural diagram of a data processing apparatus according to a thirteenth embodiment of the present application is shown.

the to-be-processed text data receiving module 1301 is configured to receive to-be-processed text data uploaded by a public cloud client, where the to-be-processed text data is obtained by text conversion of received voice data by the public cloud client;

a fifth processed text data obtaining module 1302, configured to input text data to be processed into a text post-processing model, and obtain processed text data output by the text post-processing model; the text post-processing model is obtained based on the data processing method of the first embodiment.

Optionally, the apparatus in the embodiment of the present application may further include:

and the second processed postscript data returning module is used for returning the processed postscript data to the public cloud client.

The data processing apparatus according to the embodiment of the present application is configured to implement the corresponding data processing method in the seventh method embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again. In addition, the functional implementation of each module in the data processing apparatus in the embodiment of the present application can refer to the description of the corresponding part in the seventh embodiment of the foregoing method, and is not repeated here.

Fig. 14 is a schematic structural diagram of an electronic device in a fourteenth embodiment of the application; the electronic device may include:

one or more processors 1401;

a computer-readable medium 1402, which may be configured to store one or more programs,

when the one or more programs are executed by the one or more processors, the one or more processors implement the data processing method according to any one of the first to seventh embodiments.

Fig. 15 is a hardware configuration of an electronic device according to a fifteenth embodiment of the present application; as shown in fig. 15, the hardware structure of the electronic device may include: a processor 1501, a communication interface 1502, a computer-readable medium 1503, and a communication bus 1504;

wherein the processor 1501, the communication interface 1502, and the computer-readable medium 1503 communicate with each other via the communication bus 704;

alternatively, the communication interface 1502 may be an interface of a communication module, such as an interface of a GSM module;

the processor 1501 may be specifically configured to: pre-training the text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model; acquiring label-free text data and a pseudo label of the label-free text data obtained after processing the label-free text data; training and updating the pre-training model based on the label-free text data and the pseudo label to obtain an updated model; and training and updating the updated model based on the second text data training sample and the label corresponding to the second text data training sample to obtain the trained text post-processing model.

Alternatively, the processor 1501 may also be configured to: acquiring text data to be processed; inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; the text post-processing model is obtained based on the method of the first embodiment.

Alternatively, the processor 1501 may also be configured to: receiving an instruction which is input through an interface of the instant messaging application and used for indicating that input voice data are converted into text data; performing text conversion on the voice data according to the instruction to obtain text data to be processed; inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; the text post-processing model is obtained based on the method of the first embodiment.

Alternatively, the processor 1501 may also be configured to: receiving an instruction which is input and set through the all-in-one machine equipment and used for indicating that the input voice data is converted into text data; performing text conversion on the voice data according to the instruction to obtain text data to be processed; inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; the text post-processing model is obtained based on the method of the first embodiment.

Alternatively, the processor 1501 may also be configured to: receiving voice data uploaded by a public cloud client;

performing text conversion on the voice data to obtain text data to be processed; inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; the text post-processing model is obtained based on the method of the first embodiment.

Alternatively, the processor 1501 may also be configured to: receiving to-be-processed text data uploaded by a public cloud client, wherein the to-be-processed text data is obtained after the public cloud client performs text conversion on the received voice data; inputting text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; the text post-processing model is obtained based on the method of the first embodiment.

The Processor 701 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The computer-readable medium 703 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Read Only Memory (EPROM), an electrically Erasable Read Only Memory (EEPROM), and the like.

In particular, according to embodiments of the present application, the processes described above with reference to the flow charts may be implemented as computer software programs. For example, embodiments of the present application include a computer program product comprising a computer program embodied on a computer-readable medium, the computer program comprising program code configured to perform the method illustrated by the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication section, and/or installed from a removable medium. The computer program, when executed by a Central Processing Unit (CPU), performs the above-described functions defined in the method of the present application. It should be noted that the computer readable medium of the present application can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access storage media (RAM), a read-only storage media (ROM), an erasable programmable read-only storage media (EPROM or flash memory), an optical fiber, a portable compact disc read-only storage media (CD-ROM), an optical storage media piece, a magnetic storage media piece, or any suitable combination of the foregoing. In the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In this application, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

Computer program code configured to carry out operations for the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may operate over any of a variety of networks: including a Local Area Network (LAN) or a Wide Area Network (WAN) -to the user's computer, or the connection may be made to an external computer (for example, through the internet using an internet service provider).

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions configured for implementing the specified logical function(s). In the above embodiments, specific precedence relationships are provided, but these precedence relationships are only exemplary, and in particular implementations, the steps may be fewer, more, or the execution order may be modified. That is, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules described in the embodiments of the present application may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor comprises a model pre-training module, a label-free text data and pseudo label obtaining module, a first training updating module and a second training updating module. For example, the model pre-training module may also be described as a module that performs pre-training on the text post-processing model by using the first text data training sample and the label corresponding to the first text data training sample to obtain the pre-training model.

As another aspect, the present application also provides a computer-readable medium on which a computer program is stored, the program, when executed by a processor, implementing the data processing method as described in any of the first to seventh embodiments.

In another aspect, the present application further provides a computer program, which contains computer-executable instructions, and when executed, the computer-executable instructions implement the data processing method described in any one of the first to seventh embodiments. In the embodiment of the present application, the computer program may include an APP, an applet, and the like.

The expressions "first", "second", "first" or "second" used in various embodiments of the present disclosure may modify various components regardless of order and/or importance, but these expressions do not limit the respective components. The above description is only configured for the purpose of distinguishing elements from other elements. For example, the first user equipment and the second user equipment represent different user equipment, although both are user equipment. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of the present disclosure.

When an element (e.g., a first element) is referred to as being "coupled" (operably or communicatively) with "or" connected "(operably or communicatively) to" another element (e.g., a second element) or "connected" to another element (e.g., a second element), it is understood that the one element is directly connected to the other element or the one element is indirectly connected to the other element via yet another element (e.g., a third element). In contrast, it is understood that when an element (e.g., a first element) is referred to as being "directly connected" or "directly coupled" to another element (a second element), no element (e.g., a third element) is interposed therebetween.

The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the invention herein disclosed is not limited to the particular combination of features described above, but also encompasses other arrangements formed by any combination of the above features or their equivalents without departing from the spirit of the invention. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.

Claims

1. A method of data processing, the method comprising:

pre-training a text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model;

and training and updating the updated model based on a second text data training sample and a label corresponding to the second text data training sample to obtain a trained text post-processing model.

2. The method of claim 1, wherein before the pre-training the text post-processing model using the first text data training sample and the label corresponding to the first text data training sample to obtain the pre-training model, the method further comprises:

acquiring standard text data from a standard text database, and generating corresponding non-smooth text data by adopting a preset rule;

and taking the non-smooth text data as a first text data training sample, and taking the standard text data as a label corresponding to the first text data training sample.

3. The method according to claim 1 or 2, wherein the obtaining of the unlabeled text data and the pseudo label of the unlabeled text data obtained by processing the unlabeled text data includes:

acquiring label-free text data;

4. The method of claim 3, wherein the obtaining unlabeled text data comprises:

acquiring voice data to be recognized;

and identifying the voice data to be identified by adopting an automatic voice identification technology to obtain the non-tag text data.

5. The method of claim 3, wherein the performing label prediction on the unlabeled text data by using a label prediction model to obtain a pseudo label of the unlabeled text data comprises:

training and updating the pre-training model based on a third text data training sample and a label corresponding to the third text data training sample to obtain a label prediction model;

and performing label prediction on the label-free text data by adopting the label prediction model to obtain a pseudo label of the label-free text data.

6. The method of claim 3, wherein the performing label prediction on the unlabeled text data by using a label prediction model to obtain a pseudo label of the unlabeled text data comprises:

acquiring a pre-constructed initial label prediction model; the number of network layers included in the label prediction model is greater than that included in the text post-processing model, and/or the dimensionality of each network layer in the label prediction model is greater than that of each network layer in the text post-processing model;

7. A method of data processing, the method comprising:

acquiring text data to be processed;

inputting the text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

8. The method of claim 7, wherein the obtaining text data to be processed comprises:

acquiring text data of online log reflux as text data to be processed;

after the obtaining of the processed text data output by the text post-processing model, the method further comprises:

training and updating the updated model according to any one of claims 1 to 6 based on the text data to be processed and the processed text data to obtain a transition model;

training and updating the transition model based on a sixth text data training sample and a label corresponding to the sixth text data training sample to obtain a hot standby model;

respectively calculating the accuracy of the hot standby model and the accuracy of the text post-processing model;

and when the accuracy of the text post-processing model is lower than that of the hot standby model, adopting the hot standby model as a new text post-processing model to perform the next text post-processing operation.

9. The method of claim 8, wherein after said separately calculating the accuracy of the hotstandby model and the text post-processing model, the method further comprises:

and when the accuracy of the hot standby model is lower than that of the text post-processing model, performing the next text post-processing operation by using the text post-processing model.

10. The method of claim 8 or 9, wherein said separately calculating an accuracy of said hot standby model and said text post-processing model comprises:

11. A method of data processing, the method comprising:

12. A method of data processing, the method comprising:

receiving an instruction which is input and set through the all-in-one machine equipment and used for indicating that the input voice data is converted into text data;

inputting the text data to be processed into a text post-processing model, and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1 to 6.

13. A method of data processing, the method comprising:

receiving voice data uploaded by a public cloud client;

14. The method of claim 13, further comprising:

and returning the processed text data to the public cloud client.

15. A method of data processing, the method comprising:

16. The method of claim 15, further comprising:

and returning the processed text data to the public cloud client.

17. A data processing apparatus, the apparatus comprising:

the model pre-training module is used for pre-training the text post-processing model by adopting a first text data training sample and a label corresponding to the first text data training sample to obtain a pre-training model;

the non-tag text data and pseudo tag acquisition module is used for acquiring non-tag text data and pseudo tags of the non-tag text data obtained after the non-tag text data is processed;

and the second training updating module is used for training and updating the updated model based on a second text data training sample and a label corresponding to the second text data training sample to obtain a trained text post-processing model.

18. A data processing apparatus, the apparatus comprising:

the first processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

19. A data processing apparatus, the apparatus comprising:

the second processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

20. A data processing apparatus, the apparatus comprising:

the third processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

21. A data processing apparatus, the apparatus comprising:

the fourth processed text data acquisition module is used for inputting the text data to be processed into a text post-processing model and acquiring processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

22. The apparatus of claim 21, the apparatus further comprising:

and the first processed text data returning module is used for returning the processed text data to the public cloud client.

23. A data processing apparatus, the apparatus comprising:

a fifth processed text data obtaining module, configured to input the text data to be processed into a text post-processing model, and obtain processed text data output by the text post-processing model; wherein the text post-processing model is obtained based on the method of any one of claims 1-6.

24. The apparatus of claim 23, the apparatus further comprising:

and the second processed text data returning module is used for returning the processed text data to the public cloud client.

25. An electronic device, comprising: a processor; and a memory configured to store computer-executable instructions that, when executed, cause the processor to implement the method of any of claims 1 to 6 above, or the method of any of claims 7 to 16.

26. A storage medium storing computer-executable instructions which, when executed, implement the method of any of claims 1 to 6 or the method of any of claims 7 to 16.

27. A computer program comprising computer executable instructions which when executed perform the method of any one of claims 1 to 6 or the method of any one of claims 7 to 16.